Home Research Papers Datasets State of the Art Pricing

Discover, visualize, and connect AI research papers. Explore the latest trends and insights in artificial intelligence research.

Product

Home
Research Papers
About

Support

Contact
Terms of Service
Privacy Policy

© 2026 Papersgraph. All rights reserved.

computer-vision-10

Semi-Supervised Semantic Segmentation

3260 papers • 126 benchmarks • 313 datasets

Models that are trained with a small number of labeled examples and a large number of unlabeled examples and whose aim is to learn to segment an image (i.e. assign a class to every pixel).

(Image credit: Papersgraph)

Benchmarks

These leaderboards are used to track progress in semi-supervised-semantic-segmentation-10

Trend

Dataset

Best Model

Actions

Pascal VOC 2012 12.5% labeled

Pascal VOC 2012 12.5% labeled

Cityscapes 12.5% labeled

Cityscapes 12.5% labeled

Libraries

i

Use these libraries to find semi-supervised-semantic-segmentation-10 models and implementations

HiLab-git/SSL4MIS

3 papers 1,980

Datasets

Cityscapes

nuScenes

ADE20K

SemanticKITTI

2D-3D-S

WoodScape

Subtasks

No subtasks available.

Most implemented papers

Adversarial Learning for Semi-supervised Semantic Segmentation

Yi-Hsuan Tsai, Ming-Hsuan Yang, Yen-Yu Lin, Wei-Chih Hung, Yan-Ting Liou•Wed Feb 14 2018

It is shown that the proposed discriminator can be used to improve semantic segmentation accuracy by coupling the adversarial loss with the standard cross entropy loss of the proposed model.

600

Content

Introduction Benchmarks Datasets Subtasks Libraries Papers

Cityscapes 25% labeled

Cityscapes 25% labeled

PASCAL VOC 2012 25% labeled

PASCAL VOC 2012 25% labeled

Cityscapes 50% labeled

Cityscapes 50% labeled

Pascal VOC 2012 5% labeled

Pascal VOC 2012 5% labeled

PASCAL VOC 2012 50%

PASCAL VOC 2012 50%

Pascal VOC 2012 6.25% labeled

Pascal VOC 2012 6.25% labeled

Cityscapes 100 samples labeled

Cityscapes 100 samples labeled

Pascal VOC 2012 2% labeled

Pascal VOC 2012 2% labeled

Cityscapes 6.25% labeled

Cityscapes 6.25% labeled

PASCAL VOC 2012 1464 labels

PASCAL VOC 2012 1464 labels

ScribbleKITTI

ScribbleKITTI

nuScenes

nuScenes

SemanticKITTI

SemanticKITTI

PASCAL VOC 2012 92 labeled

PASCAL VOC 2012 92 labeled

PASCAL VOC 2012 183 labeled

PASCAL VOC 2012 183 labeled

PASCAL VOC 2012 366 labeled

PASCAL VOC 2012 366 labeled

PASCAL VOC 2012 732 labeled

PASCAL VOC 2012 732 labeled

Pascal VOC 2012 1% labeled

Pascal VOC 2012 1% labeled

COCO 1/512 labeled

COCO 1/512 labeled

COCO 1/256 labeled

COCO 1/256 labeled

COCO 1/128 labeled

COCO 1/128 labeled

COCO 1/64 labeled

COCO 1/64 labeled

COCO 1/32 labeled

COCO 1/32 labeled

ADE20K 1/32 labeled

ADE20K 1/32 labeled

ADE20K 1/16 labeled

ADE20K 1/16 labeled

Cityscapes 2% labeled

Cityscapes 2% labeled

Cityscapes 5% labeled

Cityscapes 5% labeled

Cityscapes 93 labeled

Cityscapes 93 labeled

PASCAL Context 12.5% labeled

PASCAL Context 12.5% labeled

PASCAL Context 25% labeled

PASCAL Context 25% labeled

Stanford 2D-3D

Stanford 2D-3D

Cityscapes with extra (no coarse labels)

Cityscapes with extra (no coarse labels)

WoodScape

WoodScape

PASCAL VOC 2012 500 labels

PASCAL VOC 2012 500 labels

PASCAL VOC 2012 1000 labels

PASCAL VOC 2012 1000 labels

2D-3D-S

2D-3D-S

Kvasir-Instrument

Kvasir-Instrument

2017 Robotic Instrument Segmentation Challenge

2017 Robotic Instrument Segmentation Challenge

KiTS19

KiTS19

PASCAL VOC 2012 331 labeled

PASCAL VOC 2012 331 labeled

SUIM

SUIM

Cityscapes 10% labeled

Cityscapes 10% labeled

Pascal VOC 2012 50% labeled

Pascal VOC 2012 50% labeled

2 papers 28

ImageNet-S

ImageNet-S

SUIM

2017 Robotic Instrument Segmentation Challenge

2017 Robotic Instrument Segmentation Challenge

ScribbleKITTI

0

Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results

Antti Tarvainen, Harri Valpola•Thu Feb 16 2017

The recently proposed Temporal Ensembling has achieved state-of-the-art results in several semi-supervised learning benchmarks, but it becomes unwieldy when learning large datasets, so Mean Teacher, a method that averages model weights instead of label predictions, is proposed.

5071 0

Semi-supervised semantic segmentation needs strong, varied perturbations

S. Laine, Timo Aila, G. Finlayson, Geoff French, Michal Mackiewicz•Tue Jun 04 2019

This work finds that adapted variants of the recently proposed CutOut and CutMix augmentation techniques yield state-of-the-art semi-supervised semantic segmentation results in standard datasets.

356 0

Weakly- and Semi-Supervised Learning of a DCNN for Semantic Image Segmentation

K. Murphy, Liang-Chieh Chen, A. Yuille, G. Papandreou•Sun Feb 08 2015

Expectation-Maximization (EM) methods for semantic image segmentation model training under weakly supervised and semi-supervised settings are developed and extensive experimental evaluation shows that the proposed techniques can learn models delivering competitive results on the challenging PASCAL VOC 2012 image segmentsation benchmark, while requiring significantly less annotation effort.

412 0

Semi-Supervised Semantic Segmentation With Cross-Consistency Training

Yassine Ouali, C. Hudelot, Myriam Tami•Wed Mar 18 2020

This work observes that for semantic segmentation, the low-density regions are more apparent within the hidden representations than within the inputs, and proposes cross-consistency training, where an invariance of the predictions is enforced over different perturbations applied to the outputs of the encoder.

890 0

Decoupled Deep Neural Network for Semi-supervised Semantic Segmentation

Hyeonwoo Noh, Seunghoon Hong, Bohyung Han•Mon Jun 15 2015

The decoupled architecture enables the algorithm to learn classification and segmentation networks separately based on the training data with image-level and pixel-wise class labels, respectively, and facilitates to reduce search space for segmentation effectively by exploiting class-specific activation maps obtained from bridging layers.

344 0

Fast Online Object Tracking and Segmentation: A Unifying Approach

Philip H. S. Torr, Li Zhang, Luca Bertinetto, Qiang Wang, Weiming Hu•Tue Dec 11 2018

This method improves the offline training procedure of popular fully-convolutional Siamese approaches for object tracking by augmenting their loss with a binary segmentation task, and operates online, producing class-agnostic object segmentation masks and rotated bounding boxes at 55 frames per second.

1295 0

Part-aware Prototype Network for Few-shot Semantic Segmentation

Xuming He, Songyang Zhang, Yongfei Liu, Xiangyi Zhang•Sun Jul 12 2020

A novel few-shot semantic segmentation framework based on the prototype representation, capable of capturing diverse and fine-grained object features, and a novel graph neural network model to generate and enhance the proposed part-aware prototypes based on labeled and unlabeled images.

386 0

Semi-Supervised Semantic Segmentation with Cross Pseudo Supervision

Jingdong Wang, Gang Zeng, Xiaokang Chen, Yuhui Yuan•Mon May 31 2021

This paper proposes a novel consistency regularization approach, called cross pseudo supervision (CPS), which imposes the consistency on two segmentation networks perturbed with different initialization for the same input image.

1033 0

ClassMix: Segmentation-Based Data Augmentation for Semi-Supervised Learning

Viktor Olsson, Wilhelm Tranheden, Juliano Pinto, L. Svensson•Tue Jul 14 2020

This work proposes a novel data augmentation mechanism called ClassMix, which generates augmentations by mixing unlabelled samples, by leveraging on the network’s predictions for respecting object boundaries, and attains state-of-the-art results.

396 0

Adding a benchmark result helps the community track progress.

Semi-Supervised Semantic Segmentation | State-of-the-Art